An Application of DSD with Recursive Partitioning Scheme to Constant Temperature Power Sensors

摘要

The DSD (Dynamic System Doctor) is system independent, state/parameter estimation software. The DSD is based on the modeling of system evolution in terms of probability of transitions within user specified time intervals between sets of user defined parameter/state variable magnitude intervals that partition the system state space. Recently a recursive partitioning scheme has been developed for DSD that reduces the estimation time and memory requirements, as well as making DSD more user friendly. The scheme is illustrated using a non-linear model for the constant temperature power sensor, proposed for direct core power distribution monitoring in Generation 4 nuclear power reactors.